In this paper, we demonstrate how informative dialect recognition systems such as acoustic pronunciation model (APM) help speech scientists locate and analyze phonetic rules efficiently. In particular, we analyze dialect-specific characteristics automatically learned from APM across two American English dialects. We show that unsupervised rule retrieval performs similarly to supervised retrieva...

Although first demonstrators for speech to speech translation have been presented, they are still ‘show cases’ with lacking robustness for everyday usage. This lacking robustness is caused by the insufficient performance of the building blocks of speech to speech translation systems i.e. of speech recognition, spoken language translation and speech synthesis. In the context of a phonetic confer...

The problem addressed is automatically detecting, measuring and correcting nonnative pronunciation characteristics (so-called "for-eign accents") in foreign language speech. Systemic, structural and realizational differences between L1 (native language) and L2 (target language) appear as phone insertions, deletions and substitutions. A bilingual phone recognizer using native-trained acoustic mo...

Aiuruetê is a high-quality concatenative TTS system for Brazilian Portuguese. Its name (pronounced [!"#$%#&'(&]) illustrates the challenges we have fixed as a research paradigm: to feed the system with the specificities of our language, highlighted by an up-to-date discussion of the Phonology/Phonetics and prosody/segments interfaces, without a huge computational cost. The choice for the concat...

We examine the phonetic characteristics of yer and non-yer vowels in Slovak in an effort to improve our understanding of the link between phonological differences and their phonetic realizations. We test the wide-spread assumption of phonological analyses that yer vowels are phonetically identical to their non-yer counterparts with measures of vowel duration, vowel quality and the patterns of c...

Just over fifty years ago, Lisker and Abramson proposed a straightforward measure of acoustic differences among stop consonants of different voicing categories, voice onset time (VOT). Since that time, hundreds of studies have used this method. Here, we review the original definition of VOT, propose some extensions to the definition, and discuss some problematic cases. We propose a set of terms...

That talkers produce optical as well as acoustic speech signals, and that perceivers process both types of signals has become well known. Although perceptual effects due to audiovisual speech integration have been a focus of research involving the visual speech stimulus, relatively little is known about visual-only speech perception and optical phonetic signals. This knowledge is needed to expl...

The bandwidth for telephony is generally defined to be from 300–3400 Hz. This bandwidth restriction has a noticeable effect on speech quality. We present an algorithm which recovers the missing highband parts from telephone speech. We describe an MMSE estimator using hard/soft-classification to create the missing highband spectrum envelope. The classification is motivated by acoustic phonetics:...

This study investigated the use of forced alignment for automatic detection of “g-dropping” in American English (e.g., walkin'). Two acoustic models were trained, one for -in' and the other for -ing. The models were added to the Penn Phonetics Lab Forced Aligner, and forced alignment will choose the more probable pronunciation from the two alternatives. The agreement rates between the forced al...

This paper describes our recent work on developing a large vocabulary speech database for Cantonese. As a major Chinese dialect, Cantonese is spoken by tens of millions of people in Southern China and Hong Kong. It is very di erent from Mandarin or Putonghua in phonology, phonetics, vocabulary and grammatical structure. A speech database specially designed for Cantonese is urgently needed for t...